Skip to content Skip to footer
NStarX Inc.
Get in Touch
NStarX Inc.
Let's Talk
Close

Scaling AI Solutions for ISRM Needs

January 17, 2025

Business Challenge

Situation

Handling complex security queries and maintaining consistent Information Security and Risk Management (ISRM) responses across a workforce of 140,000 employees was a critical challenge for a leading healthcare and medical device manufacturer. The existing processes relied heavily on Subject Matter Experts (SMEs) and manual efforts, which were not scalable.

Challenges

Data Volume

Only 15-20% of records were matched automatically, leaving 80% to be processed manually.

Consistency Needs

Providing uniform responses across geographies, divisions, and regulatory landscapes.

Scalability

Addressing routine and complex security inquiries efficiently for a global workforce.

Knowledge Transfer

Reducing dependency on SMEs to ensure efficient and automated knowledge dissemination.

As the volume of records continued to grow, it became imperative to adopt a forward-thinking solution to improve data matching accuracy, reduce manual intervention, and ensure scalability for future demands.

Task

The primary goal was to develop an AI-powered interactive chatbot capable of:

Handling

routine and advanced ISRM-related queries with precision.

Scaling

to meet the demands of the organization’s large global workforce.

Delivering

secure, citation-supported responses aligned with ISRM policies.

Solution

A scalable, RAG-based chatbot was built using the Gen AI Lab’s MultiModal Chatbot. Features like session handling, embedding customization, and natural language disambiguation were integrated. The system indexed ISRM policies, enabling precise, citation-backed responses. A combination of structured query support and robust governance improved usability.

Technology & Tools

Infrastructure:
Cloud Kubernetes, Python, ReactJS

Frameworks:
LangChain, RAG Studio

Capabilities:
Persona customization and detailed response evaluation

Business Outcomes

Availability:
Achieved 99.98% uptime for ISRM responses.

Responsiveness:
Reduced query resolution latency by 40%.

Engagement:
Improved user satisfaction by 35%.